proxydom

CS student, Italy 🇮🇹
Passion: ethical hacking, with a focus on web applications and IoT devices (especially routers).
Github:

[proxydom @proxy_dom]

@bahrambiz @raunak_yadush @grok

[proxydom @proxy_dom]

@MetatronicDave @HackingButLegal @elder_plinius not everyone has a good gpu that can run a Haiku/Sonnet level LLM

[proxydom @proxy_dom]

@MetatronicDave @HackingButLegal @elder_plinius imagine thinking everyone has a good GPU to run local models. for example, on my laptop i can run things like qwen 20b, but i could never run something way better like gpt oss 120b or something on at least Sonnet level

[proxydom @proxy_dom]

@claudifytech straight slop lmao there are 100 free repos that do this be fr gang😭

[proxydom @proxy_dom]

@TheRabbitPy @h4x0r_dz yeah that's what I've been saying lol

[proxydom @proxy_dom]

@TheRabbitPy @h4x0r_dz i tried multiple LLMs. they tend to hallucinate if you're not "helping" them. giving them a codebase and say "ok find bugs" is NOT gonna work. and as i said, most of the times, it finds fake exploit that can't be exploited, or require particular configs to get actually exploited

[proxydom @proxy_dom]

@TheRabbitPy @h4x0r_dz most people put claude to look at code, and don't even bother to check if it's actually right. most of the times the LPE/RCE the LLM found is probably a false positive, like a null pointer (even tho it's never null)

[proxydom @proxy_dom]

@S1r1u5_ too bad AI tech bros can't read, they're still gonna blindly believe that IT jobs (and cybersec one) are cooked because of AI🫤

[proxydom @proxy_dom]

@h4x0r_dz that bounty was promised to them 3000 years ago!

[proxydom @proxy_dom]

@stalmico @h4x0r_dz using AI to leak personal data, or just to execute code depending on how it was setted up

[proxydom @proxy_dom]

@thedawgyg yeah at this point just sell them lmao

[proxydom @proxy_dom]

@vxunderground yeah most people believe anything they see on social media, without even checking if it's true bruh🫩

[proxydom @proxy_dom]

@With_Cornelius @vxunderground holy opsec larping bro, bro done got AI advice on ts😭

[proxydom @proxy_dom]

@h4x0r_dz weird, i never had to do this?

[proxydom @proxy_dom]

@S1r1u5_ completely wrong. what if I'm a red teamer with lecit intentions? and nobody would use kyc, they would lose a lot of customers and people would probably switch to other models (like Kimi, honestly underrated)

[proxydom @proxy_dom]

@thedawgyg makes sense. did you use sonnet or opus? i heard opus is the smartest model rn, and some people got it to write exploit code and find 0day, like someone here on X found a 0day in the windows kernel. truly fascinating lol

[proxydom @proxy_dom]

@realcarlgrant @VladTheInflator @grok bro just read you don't need ai to summarize it💔💔💔

[proxydom @proxy_dom]

@bhavika_jangid_ @BasedDaedalus

[proxydom @proxy_dom]

@dmcxblue @h4x0r_dz it is. an app shouldn't automatically open things without notifying the user. while it isn't that much of an issue since almost no one uses Whatsapp desktop AND has python installed, it's still a "vulnerability" (if it wasn't, telegram wouldn't fix it)